Fluid/Solid Boundary Conditions in Non-Isothermal Systems
Total Page:16
File Type:pdf, Size:1020Kb
Fluid/Solid Boundary Conditions in Non-Isothermal Systems Ph DANIEL E. ROSNER NASA Grant No. NAG 3-1951 Yale University, Department of Chemical Engineering High Temperature Chemical Reaction Engineering (HTCRE-) Laboratory New Haven, CT 06520-8286 USA 1. Introduction, background and objectives In a short paper (Phys. Fluids A 1 (11) 1761-1763 (1989)) the PI pointed out for the first time that vapor-filled nonisothermal ampoules operating in micro-gravity will experience convection driven by both "wall-induced thermal creep"* as well as thermal stresses (in the bulk vapor). But, as emphasized in our subsequent review of thermally-induced particle phoresis (Rosner et.aL, 1992) these "thermal creep" boundary condition (BC-) phenomena also determine the readily observed migration rates ("phoresis") of suspended "small" solid particles in suspension. This led us to the idea that careful phoresis measurements in a suitably designed microgravity environment (to preclude complications such as 'sedimentation' for particles large enough both to sustain large temperature differences and avoid appreciable Brownian motion) could be used to obtain unambiguous information about the nature of the tangential momentum transfer boundary condition over a broad range of Newtonian fluid densities. This would be especially interesting since present theoretical methods are really valid only in the low density ("ideal"-) gas limit. As noted below, we recently carried out numerical simulations of the "low-density" Boltzmann equation which not only clarified the wall ("creep") BC, but also thermal stress (non-Newton-Stokes-) effects within such crystal- growth ampoule flows. However, to clarify the nature of the BC transition from the Enskog- Chapman regime to the technologically important but theoretically less tractable case of liquid- like densities (including supercritical polyatomic vapors) it is inevitable that high-quality experimental data for carefully selected fluid/solid interfaces will be required. The best environment for obtaining these data appears to be the microgravity environment, as spelled out in Section 2 below. Previous theoretical work on "thermal creep" at fluid/solid interfaces (going back to J.C. Maxwell[1879]) has been understandably focused on "structureless", single component low density gas motion adjacent to a non-isothermal smooth flat solid surface, usually exploiting an approximate (eg. linearized) model of the Boltzmann equation and a "diffuse reflection" molecule/surface interaction model. However, since many applications are multi-dimensional and involve non-dilute mixtures of disparate mass polyatomic molecules, frequently well outside of the domain of "ideal" gas behavior, many generalizations are urgently needed to understand and ultimately predict such flows, whether in micro-gravity or in ground-based laboratories. Progress will be slow in the absence of reliable experimental data on the 591 dimensionlessthermal creep (or wall 'slip') coefficient, Ctc , appearing in the relevant tangential momentum BC: (Vt,w)tc =Ctc'n'(°lnTnuid,w/'X) ( 1 ) Here (vt, w )tc is the thermal creep contribution to the tangential component (x-direction) of the *It is rather remarkable that, despite the small Knudsen numbers typically associated with these flows, one cannot apply "no-slip" boundary conditions along the non-isothermal ampoule walls-ie, the walls drive a significant amount of convection ! fluid velocity relative to the solid surface, n is the momentum diffusivity ("kinematic viscosity", m/r ) of the fluid, and Tfluid, w is the fluid temperature evaluated at the fluid/solid interface. One sees from Eq.(1) that such wall-induced creep flow is subtle since for solids in gases, if Ctc =O(1) (as appears to be the case from the limited available experiments) then the Mach number associated with the wall creep flow is of the same order of magnitude as the fractional change of gas temperature over one mean-free-path. In our view, the ultimate goal of this line of research is a rational molecular level theory which predicts the dependence of Ctc on relevant dimensionless parameters describing the way fluid molecules interact with the solid surface, and how they interact among themselves (see, eg., Rosner and Papadopoulos,1996). Accordingly, microgravity-based experimental data (Section 2 below) will be an important step forward. 2. Photophoretic "space race" experiments It might be thought that the best way to experimentally obtain this rather important dimensionless coefficient would be the "direct" one of measuring fluid velocities near non- isothermal solid surfaces using, say, laser velocimetry on very small (negligible inertia) particles immersed in the local fluid. However, even if the microgravity environment were used to preclude any confusion from bouyancy-driven convection, the phenomenon of thermo- phoresis (eg., Rosner et.al., 1992, Gomez and Rosner,1993) would inevitably cause the local motion of small "tracer" particles to depart systematically from that of the non-isothermal host fluid! To avoid this dilemma we suggest exploiting the closely related phenomenon of photophoresis under microgravity conditions. This phenomenon (in which the intraparticle temperature nonuniformity which causes the thermal creep in the surrounding fluid is induced by the absorption of incident radiation) is quite sensitive to the momentum transfer boundary condition of interest (Eq.(1)) and becomes rather important at particle sizes at which sedimentation phenomena would normally set in for ground-based experiments (Castillo et.al.,1990, Mackowski 1989,1990). Thus, microgravity experiments involving deliberately suspended supermicron spherical solid particles of low thermal conductivity but with known "optical" properties (over a wide range of transparent fluid densities) could prove to be 592 extremely valuable in defining the nature of the momentumtransfer boundary condition "transition" (ie.,Ctc-trends)from thelow density(idealgas-)limit to the bona-fide "liquid" limit. One simple realization of this basic idea, would, in effect, be a "space-race" of illuminated, absorbing spheres, each immersed and simultaneously released in a different pure, transparent Newtonian fluid (including dense as well as low density polyatomic gases, and certain transparent Newtonian liquids). These low thermal conductivity spheres would be illuminated by a common non-polarized, omni-directional radiation source, and each would be large enough to sustain a significant temperature difference across its diameter (despite the isothermal environment far from each sphere). Moreover, as mentioned above, in the microgravity environment 'sedimentation' or 'bouyant rise' would be eliminated as a complication, broadening considerably the choice of informative fluid/solid combinations. Therefore, from the "photo-phoretic" drift velocity of these spheres, (recorded "from above" via a TV camera ) and the relevant theory (see, eg., Rosner et.al.,1992, and Mackowski, 1989), it should be possible to extract the tangential momentum thermal creep coefficient , Ctc , operative at each fluid/solid interface. We have already identified most of the basic requirements for such experiments t(o extract the dimensionless thermal creep coefficient, Ctc, from measurements of photophoretic drift velocities in different Newtonian fluids). Indeed, we are currently evaluating the possible construction/use of composite test spheres to amplify the effects sought---ie., opaque thin shells containing a thermally insulating rigid "core" of small mass. As a part of this research program we are currently working out the (asymptotic) theory of photophoresis for such test-spheres. 3. Thermophoresis of isolated solid spheres and aggregates in gases A little-noticed but potentially very important feature (see below) of the existing theory of solid sphere thermophoresis in ideal gases is that for sphere thermal conductivities less than about 2.3 times that of the background gas, the sphere thermophoretic diffusivity can actually exceed that in the free-molecule limit! This remarkable feature, which translates to larger spheres moving faster than their smaller counterparts, may be partially 'rationalized' by noticing that, whereas the (Waldmann-) free-molecule limit calculation (see,eg., Rosner,1980, Garcia-Ybarra and Rosner,1989, or Loyalka,1992) does not involve either the abovementioned creep coefficient, or the sphere thermal conductivity, the near-continuum value does contain both Ctc, and sphere thermal conductivity. We mention this behavior here because, in related studies, we have been clarifiying the effects of intrinsic thermal conductivity, particle morphology, and Knudsen number on the thermophoretic properties of suspended particles (in ideal gases), and 'coagulation-aged' populations thereof (Rosner and Khalil, 1998). While a comprehensive theory for aggregate thermophoresis as a function of Knudsen number is not yet available (see,eg., Rosner et.al.,1991), and outside the scope of the present program, we have proposed that a rational estimate can be obtained by imagining that an aggregate behaves like a homogeneous sphere with an effective mobility diameter about equal to the gyration diameter but with an effective thermal conductivity equal to that expected for the 'granular medium' (of spherules) existing at the gyration radius (for some details, see Tandon and Rosner, 1995). Results of this plausible model indicate that the thermophoretic transport rates of aggregates of even